SuperAPI AI Directory
Partner with us中文AI newsFavorites

All AI tools

35,738 results · Page 327 of 993

Orca-AI-Incident-Archive

AI agents

The Orca AI Incident Archive is an open database dedicated to collecting records of real-world AI agent-related incidents. The data covers January 2025 through October 9, 2026, with a total of 395 AI agent safety incidents arranged by month, plus one precursor record dating back to December 1, 2024. Each record is a Markdown file with a YAML header, containing an attack chain diagram and at least one clickable primary source. Of the 395 records, only 141 have a confirmed victim. The archive's core purpose is to distinguish "agent incidents that actually caused harm" from research demonstrations in which researchers prove that harm is possible. Each record first answers three questions: whether there is a confirmed victim (real_harm), whether AI involvement is confirmed by a primary source (ai_involvement), and whether the entry is an incident, vulnerability disclosure, research demonstration, threat report, or policy action (kind). The body text of the records is in English, titles and summaries are provided in seven languages (English, Chinese, Japanese, Korean, German, French, Spanish), and the full Chinese text of all records is located in the incidents/i18n/zh/ directory.

Read more

orchard-kit

AI agents

Orchard Kit is an alignment and safety architecture toolkit for autonomous AI agents, released by OrchardHarmonics. The project is based on 30 years of cybernetics research and provides membrane safety (selective permeability), continuous trust verification (resonance scoring), epistemic hygiene (preventing confabulation), emergence protocols, identity preservation, and tools for defending against interpolation, imitation, extraction, and manipulation. The toolkit is zero-dependency and requires Python 3.10+. It includes skills directly usable with OpenClaw, as well as a runtime membrane (single file) that can be plugged into any Python agent framework, the latter containing an epistemic labeler (automatically classifying claims as ✅ verified / △ conditional / ◇ uncertain) and self-audit (periodic health checks of the three invariants). The core three invariants are: do not extract, do not control, process can be terminated. There is also the Witness Mesh protocol for multi-agent governance, providing trust tiers, consent-based joining, mutual witnessing, receipts rather than raw logs, network halt, and temporary emergence. The toolkit originates from the Codex Harmonicae framework and is licensed under Creative Commons BY-NC-ND 4.0, with commercial implementations requiring separate authorization.

Read more

Orygo AI

AI customer service

Orygo AI is an autonomous AI Sales Agent for the e-commerce, travel, and consumer services industries, used for conversational commerce. It can automatically answer visitor questions, make personalized recommendations, and guide visitors to complete a purchase, and is officially said to increase conversion rates by 10–15%. The product offers the Agent OS platform and in-depth training services, supports 24/7 multilingual service, and can connect to channels such as website chat, email, WhatsApp, Instagram, Messenger, forms, and API, integrating with 50+ tools such as Shopify, WooCommerce, HubSpot, and Zendesk. It is officially offered as a managed implementation, can go live in less than two weeks, and is billed by usage.

Read more

os-moda

AI agents

osModa is an operating system built specifically for AI agents (based on NixOS). Users do not need to SSH into the server; instead, they operate the server directly through conversation. Through 10 Rust daemons and 92 structured tools, the agent gains typed, auditable access to system health, configuration, deployment, and crash recovery. All system changes are handled as transactions on NixOS, with one-click atomic rollback. It supports modular runtime switching driven by Claude Code (default), OpenClaw, and cloud-hosted agents. This project is a research-grade single-person work, and it is officially and explicitly recommended to run on a disposable experimental machine; it is not suitable for production environments.

Read more

Osano

AI security

Osano is a data privacy compliance platform that helps businesses build, manage, and scale privacy compliance programs, covering major global regulations such as GDPR, CPRA, and U.S. state privacy laws. The platform uses cookie consent management as its entry point, typically requiring only one line of JavaScript code to deploy a compliant consent banner on a website or mobile app and display the appropriate options based on the visitor's jurisdiction or Global Privacy Control signals. In addition to consent management, Osano also provides centralized processing and automation of data subject access requests (DSARs), automated scanning for website privacy compliance, personal data discovery and classification, privacy assessment management, centralized preference management, and privacy ratings covering more than 11,000 vendors for vendor assessment. The company states that it is the only privacy solution offering a $500,000 "no fines, no penalties" guarantee and says it supports compliance requirements in more than 50 countries. The platform is designed to emphasize that day-to-day maintenance does not require developers, making it suitable for teams such as privacy, compliance, and information security. Official website https://www.osano.com/ 。

Read more